Crystalline semiconductor shatters century-old thermoelectric limit

A collaboration led by the Jawaharlal Nehru Centre for Advanced Scientific Research (JNCASR), with partners at the University of Sydney and the Indian Institute of Science (IISc), has demonstrated a crystalline semiconductor whose thermoelectric response exceeds a century-old theoretical ceiling by a factor of nearly a thousand.

The team grew thin films of scandium nitride (ScN) on magnesium oxide substrates using ultrahigh-vacuum magnetron sputtering. By deliberately doping the films with magnesium, they compensated for the material's natural free electrons, creating what they call a heavily doped, highly compensated (HDHC) semiconductor — a crystal where positive and negative dopant atoms are randomly scattered in nearly equal numbers.

X-ray diffraction and atomic-resolution electron microscopy confirmed the films remained single-crystalline and epitaxial, with no secondary phases or precipitates. Yet in these structurally perfect films, the measured Seebeck coefficient — the voltage generated per degree of temperature difference — reached values several hundred to over a thousand times larger than typical inorganic semiconductors, and nearly a hundred times beyond the previously accepted limit for solids.

In a roughly 200-nanometre-thick film, the team recorded a Seebeck coefficient exceeding –124.6 millivolts per Kelvin near room temperature. That places the material firmly in the range previously associated only with liquid electrolytes and ionic conductors such as hydrogels. The effect grew stronger as the films were made thinner.

"We were not looking for a record-breaking number when we started this work," said Prof. Bivas Saha, who led the study. "What we found instead is that a fully crystalline, epitaxial, single-phase semiconductor can behave thermoelectrically like a liquid electrolyte."

The team also built a prototype photon sensor using an HDHC ScN film with chromium contacts. Illuminating one contact with a laser created a localized temperature difference, producing a Seebeck response of –102.4 millivolts per Kelvin — large enough for classical light detection and, with further optimization, potentially for single-photon-level detection near room temperature. The response was fast and repeatable, and an Indian patent application has been filed.

The findings, published in Science, point to applications in ultrasensitive temperature sensing, low-noise thermal imaging, heat-flux detection, bolometric devices, and cryogenic thermoelectric single-photon detectors for quantum technologies. The steep temperature dependence of the effect could also enable temperature-controlled Seebeck switches.